According to Chinese Soil Taxonomy(CST), we selected 16 soil diagnostic horizons and diagnostic characteristics, which can be used to classify cambisols at different levels based on 97 cambisol soil profiles in Soil Series of China (Henan), and we built a data matrix with different characteristics and hierarchy weights of these indexes. Then we extracted 6 centroid soil families by fuzzy k-means algorithm model, and calculated the taxonomic distance between sampled and centroid soil families, meanwhile numerically classified the 97 cambisol soil profiles. We find that the rationality of the traditional hierarchic CST can be evaluated according to the consistence of traditional hierarchic classifying units and the numerical soil classification derived from the taxonomic distance between sampled and centroid soil families.